Docker Python application

#docker   #python   #app   #devcontainer  

---

Table of Content

Docker Python application

Using pysimplegui to build simple python GUI application

Using Docker as production, development and test environment Using VSCode devcontainer as development setup

Dockerfile

Build from four stage: 1. python_base: base on ubuntu 22.04 that add user and install python 2. deploy: Add python application dependencies install by apt and pip (without the application it self) 3. prod: Install the application using whl build by development 4. development: Add all packages need by the development and build system

---

Helper scripts

Deploy

Check that the application can installed and run

• Deploy container include all package needed by the application for running (apt and pip)
• Run docker container with user permission and x11 support
• Create share from <project>/dist folder to install the application on docker container
• Install run by the user from /dist folder using pip

run_deploy_container.sh

docker run -it --rm \
    --name py_deploy \
    --hostname deploy \
    -u=$(id -u $USER):$(id -g $USER) \
    -e DISPLAY=$DISPLAY \
    -v /tmp/.X11-unix:/tmp/.X11-unix:rw \
    -v $(pwd)/dist:/dist \
    py_gui_demo:deploy \
    /bin/bash

Prod

Create docker container that run the application

• Base on prod stage that install all application package dependencies
• Get the application version to build from outside (vscode task, command line)
• Set entrypoint

PATH

The application entry point installed in <user home>/.local/bin

The docker add this environment variable with ENV command

ENV PATH=${PATH}:/home/user/.local/bin

production stage

# ###################      production     #############################

FROM deploy as prod
ARG APP_VER=0
USER user
WORKDIR /home/user
RUN echo ${APP_VER}
COPY ./dist/py_gui_demo-${APP_VER}-py3-none-any.whl /tmp
RUN pip install /tmp/py_gui_demo-${APP_VER}-py3-none-any.whl
ENTRYPOINT ["my_app"]

run_prod_container.sh

docker run -it --rm \
    --name py_prod \
    --hostname prod \
    -u=$(id -u $USER):$(id -g $USER) \
    -e DISPLAY=$DISPLAY \
    -v /tmp/.X11-unix:/tmp/.X11-unix:rw \
    py_gui_demo:prod

Base

Create docker container that include only the ubuntu base with python and pip install without all application package dependencies, use to check whl and deb installation

• Container with user context and x11 support
• Share dist and deb_dist folder from host

run_base_container.sh

docker run -it --rm \
    --name py_base \
    --hostname base \
    -u=$(id -u $USER):$(id -g $USER) \
    -e DISPLAY=$DISPLAY \
    -v /tmp/.X11-unix:/tmp/.X11-unix:rw \
    -v $(pwd)/dist:/dist \
    -v $(pwd)/deb_dist:/deb_dist \
    py_gui_demo:base \
    /bin/bash

---

VSCode

devcontainer

tasks

• Add task build each dockerfile stage

{
    "version": "2.0.0",
    "tasks": [
        {
            "label": "build deploy container",
            "type": "shell",
            "command": "docker build --target deploy -t ${workspaceFolderBasename}:deploy -f .devcontainer/Dockerfile .",
            "problemMatcher": []
        },
        {
            "label": "build base container",
            "type": "shell",
            "command": "docker build --target python_base -t ${workspaceFolderBasename}:base -f .devcontainer/Dockerfile .",
            "problemMatcher": []
        },
        {
            "label": "build prod container",
            "type": "shell",
            "command": "docker build --target prod -t ${workspaceFolderBasename}:prod -f .devcontainer/Dockerfile --build-arg APP_VER=${input:app_version} .",
            "problemMatcher": []
        }

    ],
    "inputs": [
        {
            "id": "app_version",
            "description": "app_version",
            "options": ["0.0.1", "0.0.2"],
            "type": "pickString"
        },
    ]
}